[Cyborg] dolphin language

Joe Goldbeck joseph.goldbeck at gmail.com
Thu Dec 15 14:52:34 PST 2011


A neat experiment for sure. The cymascope, while also neat, doesn't really
seem to play any part in the actual science. As far as I can tell, it's
pretty much a dish of water with a speaker below it and a camera above it.
You definitely get some beautiful shapes out of it ( for ex,
http://youtu.be/stMCW6L5GXA ), but it seems unlikely that it's pulling out
any more useful information about the sound input than you'd get from
computing the spectrogram.

I much prefer this: http://youtu.be/Yp1wUodQgqQ

Joe

On Thu, Dec 15, 2011 at 6:32 AM, Robert Picone <rpicone at gmail.com> wrote:

> While I'd definitely say that they seem to be overstating their results,
> what they don't seem to be saying is that the dolphins produce direct
> mimicry of the sounds they perceive.  That would probably not further their
> thesis of language, and would indeed  probably be rather detectable (though
> it probably wouldn't result in the same sound twice on microphone, both
> because echolocation is highly dependent on the dolphin's frame of
> reference, and because some deal of fidelity would likely be lost in the
> process, but we don't quite know from where)
>
> Anyway, I think that what they're working off of here is the speed with
> which the dolphins associated the playbacks to the real objects when the
> system was unfamiliar to them.  Animals can be trained to identify objects
> from videos of the object, but simply showing an animal with no experience
> with video a video seems unlikely to result in identification of the
> object.  So, either dolphins have some innate sense of sonic-visual
> language, or dolphins are exceptionally good at figuring out games when
> promised rewards...  The latter is true, but they don't seem to be
> distracted by that.  A better test that seems more in line with other
> standards for studying animal communication might be to playback the
> generated data to one dolphin, then determine if that dolphin can direct a
> second dolphin to choose the correct object without seeing it.
>
> Anyway, if dolphins do have such language, there is no particular reason
> it should be easily distinguishable for us.  We don't even really know what
> features they would prioritize in communicating such information, let alone
> how it would be encoded into the sounds available to them.  Would
> communicating the difference in resonance/scattering from rubber to steel
> be of importance?  How much shape detail is necessary?  A couple
> concave/convex regions?  edges? all the detail they can perceive in a long
> series of echolocative bursts?  Is perspective of the original image
> preserved in repetition?  Is overall distance?  The possibility that
> perspective is not maintained between representations of the same object
> alone account for millions of ways to "describe" a given object in such a
> language, so a complete inability to detect any sort of language-like order
> would be very much understandable.
>
> On Wed, Dec 14, 2011 at 11:40 PM, Mitch Altman <maltman23 at hotmail.com>wrote:
>
>>  If Dolphins can communicate shapes that they haven't actually
>> experienced, but are shapes from their imagination, that implies that
>> they are capable of either design, or art, or both.  That's way cool.
>>
>> Mitch.
>>
>>
>> ------------------------
>>
>> > Date: Wed, 14 Dec 2011 23:23:48 -0500
>> > From: mrericboyd at yahoo.com
>> > To: cyborg at lists.noisebridge.net
>> > Subject: [Cyborg] dolphin language
>>
>> >
>> > http://speakdolphin.com/ResearchItems.cfm?ID=20
>> >
>> > I'm having a little difficulty understanding the article, but my gist
>> is
>> > this:
>> >
>> > Dolphins have a visual language, where they "talk shapes" with each
>> > other, by manually reproducing the sounds that they would hear if they
>> > did their ecolocation against an object. So basically, not only can
>> > they make the outgoing ultra-sonic click-train, they can *also* make
>> the
>> > incoming echo sounds, and thus "repeat" what they "heard" to each
>> > other. The article even speculates that they can generate the sounds
>> > for objects they haven't actually encountered, but only imagined - if
>> > so, you can imagine that would be super powerful!
>> >
>> > Translating such a powerful visual language into English is left as an
>> > exercise for the reader :-)
>> >
>> > The scientists claim to use a CymaScope, anyone seen one of these
>> things?
>> >
>> > http://www.cymascope.com/cymascope.html
>> >
>> > I'm super intrigued by these results, but I have to say, the article
>> > doesn't seem to actually say it the way I said it. The results all talk
>> > about just replying the reflected sounds from real objects, and having
>> > the dolphins recognize them - this is not surprising, it would be like
>> > us seeing a video of an object, and then later pointing out the object:
>> > not necessary indicative of *language*, merely of perception and
>> > memory. You could train most any mammal to do it, fairly easily. And
>> > when the article later talks about language, there is a suspicious lack
>> > of experimental details. Do the dolphins actually generate the much
>> > more complicated reflected sound waves, and not just the outgoing
>> > clicks? That would be simple to verify. But if they do generate such
>> > complicated sounds, you'd think this would have been discovered and
>> > understood long ago, since it would seem really obvious if dolphins
>> > repeated sounds they just heard back to other members of their tribe...
>> > you'd literally hear the same thing twice on a microphone, but only
>> once
>> > preceded by the click train.
>> >
>> > Or maybe that's one of those things that is only obvious in hind-site?
>> >
>> > Anyway, fascinating, and I still want to build my ultra-sonic
>> > audification echo-location rig :-)
>> >
>> > Eric
>> >
>> > _______________________________________________
>> > Cyborg mailing list
>> > Cyborg at lists.noisebridge.net
>> > https://www.noisebridge.net/mailman/listinfo/cyborg
>>
>> _______________________________________________
>> Cyborg mailing list
>> Cyborg at lists.noisebridge.net
>> https://www.noisebridge.net/mailman/listinfo/cyborg
>>
>>
>
> _______________________________________________
> Cyborg mailing list
> Cyborg at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/cyborg
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.noisebridge.net/pipermail/cyborg/attachments/20111215/686a636e/attachment.htm 


More information about the Cyborg mailing list